Overview
Titled Automated Schedules — “Manage automated tasks for this instance”. Recurring actions against this one machine.Navigation: Compute → Virtual Machines → click a machine → Schedules.
What it’s for
The empty state doubles as the explanation, offering three cases:What a row shows
Each schedule shows its Timezone and Status, with transient chipsCreating... and Deleting... while work is in flight.
Editing only changes the description
Schedules versus backup schedules
Two different things, in two different sections:The empty state mentions “Automated Backups” as a use case, but backup automation is configured in the Backups section, not here. Don’t create a schedule expecting it to produce backups.
Common task: stop development machines overnight
1
Confirm the machine is genuinely non-production
A stop schedule on something serving traffic is an outage. Check what depends on it first.
2
Create the stop schedule
Create Schedule, choosing a stop action outside working hours — 19:00 is a reasonable default for most teams.
3
Decide whether you also want a start schedule
Starting at 08:00 means the machine is warm when people arrive. Leaving it stopped means it only runs when someone actually needs it, which is cheaper. Either is defensible; pick deliberately.
4
Use one timezone across the fleet
UTC for everything. Mixed timezones make “did everything stop last night?” much harder to answer, and they shift twice a year.
5
Write the reason in the description
This is the one field you can edit later, and it’s the only place the intent is recorded. “Stop dev overnight — cost saving” beats an unexplained schedule somebody is afraid to delete.
6
Verify it fired
Check Jobs the next morning, widening the time window past the six-hour default. A schedule can exist while every run fails.
Empty and error states
Related
Backups tab
Backup automation, which lives there instead.
Jobs
Confirming a scheduled run actually happened.
Quotas and limits
What stopping a machine does and does not free.